EP1986374A1

EP1986374A1 - Gateway with improved QoS awareness

Info

Publication number: EP1986374A1
Application number: EP07107100A
Authority: EP
Inventors: Michaël BECK; Eric Borghs; Steven BOUCQUÉ; Thierry Pollet; Johan Haspeslagh
Original assignee: Interuniversitair Microelektronica Centrum vzw IMEC; Alcatel Lucent SAS
Current assignee: Interuniversitair Microelektronica Centrum vzw IMEC; Alcatel Lucent SAS
Priority date: 2007-04-27
Filing date: 2007-04-27
Publication date: 2008-10-29
Anticipated expiration: 2027-04-27
Also published as: US20080267087A1; DE602007006633D1; ATE468688T1; EP1986374B1; US8184550B2

Abstract

The present invention is related to a device and a method for exchanging data frames over a networking device. In particular, the device exchanges data between a WAN and one or more LAN segments in an optimized way leading to a better Quality of Experience for the end user. The networking device (1) comprises an interface (2) for exchanging data frames over an access network, at least a first and a second subnet interface (4,5) for exchanging data frames and arranged for being coupled to a network, a memory (6) for storing classification rules, a classification agent (7) coupled to the memory (6) and arranged for extracting information from an incoming data frame and for applying the classification rules to the extracted information in order to determine the interface via which said incoming data frame is to be forwarded, and a Quality of Service monitoring agent (9) being arranged for retrieving Quality of Service information from the subnet interfaces (4,5) and for dynamically updating at least one of the classification rules according to said QoS information.

Description

Field of the Invention

The present invention relates to a networking device for exchanging data frames. In particular, the device exchanges data between a WAN and one or more LAN segments.

State of the Art

A Residential Gateway is well known in the art. Residential Gateways are common devices used in home networking. They provide bridging and routing functions between on the one hand a public Wide Area Network (WAN) infrastructure such as the Internet, and on the other hand a Local Area Network (LAN). The LAN is essentially an in-house network, and typically contains a wired Ethernet-based segment and/or a wireless WiFi-based segment.
When multiple LAN segments are present, packets entering the Residential Gateway from the WAN side are forwarded to the respective LAN segments by application of standard bridging and/or routing rules. These rules take into account the destination address (MAC address or IP address and/or TCP/UDP port number) of the incoming packets to determine the outbound port. Additionally, the optional Virtual LAN Identifier (VID), if present, may be used to refine the forwarding decision. These very rudimentary classification rules are stored in a forwarding information base (routing table, bridging table, VLAN membership map, ...).
Once an outbound port has been selected, Quality of Service (QoS) related header fields such as the IP Differentiated Services Code Point (DSCP) field, may be checked against a set of QoS rules to determine the queue to which the packet shall be written. This behaviour is common in existing devices, as may be evidenced by way of example by the operating manual of the Adtran NetVanta products (Adtran AOS QoS Configuration Guide, 61200860L1-29.3E, March 2006, available on-line (accessed 2007-04-05) at URL http://www.adtran.com/adtranpx/Doc/0/DVA89BJ6DKAKR25MCQM8BG VD75/61200860L1-29.3E.pdf).
More sophisticated ways of classifying incoming packets are known in the art. The classifier may take into account information about source and/or destination, as stored in the incoming packet, to determine the "flow" to which the incoming packet belongs. This "flow" represents a certain application or service, and subsequent manipulations such as forwarding, tagging, and encapsulating, may be based on the identified flow to present application-aware behaviour.
A basic form of flow processing is used for demultiplexing "conversations" over the physical segments that make up an aggregated link in Ethernet's Link Aggregation. A residential gateway using conversations to demultiplex data traffic over interfaces with different transmission profiles is disclosed in patent application EP 1328091A , where it suffices to read the term "residential gateway" for the more general term "modem system".
Even more advanced classification schemes are known in the art for a router as in WO 2004/080024A and for an access node (for example, patent application EP1662718A ). Obviously, such classification schemes could also be used in a residential gateway.
It is noted that residential users often use pre-existing cabling infrastructure to deploy point-to-point or point-to-multipoint links between devices that must be capable of interacting with each other. Industry standards exist for the use of in-house telephone wiring (e.g., HomePNA (HPNA), denoting Home Phoneline Networking Association) or coaxial television wiring (e.g., MoCA - Multimedia over Coax Alliance) for this purpose. These non-native networks are cumbersome to install and maintain. As they poorly interact with dedicated LAN equipment, the result is a sub-optimal patchwork of network segments.
By their nature, the different segments of the patchwork network are more or less suited for different services. Local Area Networks offering different services over different physical interfaces of the same device are well known in the art; this paradigm is sometimes referred to as "coloured interfaces" (Beck M., "Ethernet in the First Mile: the IEEE802.3ah EFM Standard.", 1st edition, NY: McGraw-Hill Professional, 2005, p.86-87).

Aims of the Invention

The present invention aims to provide a networking device for exchanging data frames wherein the transmission of incoming data over various segments of a network is improved.

Summary of the Invention

The present invention provides a networking device. The device comprises an interface for exchanging data frames over an access network, at least a first and a second subnet interface, a classification agent and a Quality of Service (QoS) monitoring agent. The subnet interfaces are arranged for being coupled to at least a first and a second network, respectively. The memory stores classification rules applicable to the data frames. The classification agent is coupled to the memory and arranged for extracting information from an incoming data frame. Further, the classification agent is arranged for applying the classification rules to the extracted information in order to determine the interface via which the incoming data frame is to be forwarded. The Quality of Service monitoring agent is arranged to retrieve Quality of Service information from the subnet interfaces. The QoS monitoring agent is further arranged for dynamically updating at least one of the classification rules according to the retrieved QoS information.
Over an access network, different services are offered with different requirements related to e.g. capacity, latency and packet loss. Different incoming data frames are forwarded to respective subnets. Both native subnet segments and segments built on legacy cabling may achieve different data rates depending on the quality of the medium and the distance between the attached stations. A prior art analysis shows that it is hard to predict the overall capacity of a patchwork network. Data frame transmission is optimized by taking into account different QoS characteristics in terms of among other things bit rate, latency and packet loss statistics (collectively referred to as a 'profile') of the different subnets. This is provided by combining a classification agent with a QoS monitoring agent. Optimised forwarding rules are added to the standard classification rules. This leads to a better utilization of the available bandwidth and to a better Quality of Experience for the end user.
The networking device of the present invention may be a Residential Gateway. Such a gateway is typically a hardware device connecting a home network or a local area network (LAN) with a wide area network (WAN) or the internet. This comprises native LAN segments (e.g. Ethernet, WiFi and other segments known to the person skilled in the art), as well as with segments built on legacy cabling (e.g. HPNA, MoCA and other segments known to the person skilled in the art).
In a preferred embodiment, the access interface of the networking device of the present invention may be a WAN interface arranged for being coupled to an access link, in particular a PON link. In another embodiment the first and second subnet interface may be arranged for being coupled to the same network.
In another embodiment, the networking device of the present invention may function as a traffic aggregator. The networking device is connected to a video decoding unit via its access interface, using a point-to-point link technology. With the networking device of this embodiment, it is possible to extend the classification rules with rules for forwarding subscribed coded video flows to the access interface.
The information extracted from the incoming data frames may comprise a source MAC address and/or a destination MAC address; a source IP address and/or a destination IP address; a virtual LAN identifier and/or a user priority indicator. Upon entry, the classification agent extracts this information from the incoming data frames. Based on this information, the classification agent recognizes the frame as being part of a particular flow and forwards it to the most appropriate subnet. Multiple, heterogeneous PHYs of the patchwork network are taken into account to optimize the data transfer.
According to the present invention is further presented a method for exchanging data between an interface for exchanging data frames over an access network and at least a first and a second subnet interface. The method comprises the steps of: a) setting initial classification rules for identifying data frames; b) querying said subnet interfaces in order to retrieve QoS information; c) updating at least one of the initial classification rules to rules for forwarding data frames; and d) selecting an interface according to said rules for forwarding an incoming data frame in said classification agent. This method can transmit data frames in an optimized way by taking into account the different profiles of the different subnets. This is provided by combining a classification agent with a QoS monitoring agent. Optimised rules for forwarding are added to the standard classification rules. This leads to a better utilization of the available bandwidth and to a better Quality of Experience for the end user.
In another embodiment, step c) of the method of the present invention further comprises the step of extending the forwarding rules with forwarding rules that can forward subscribed coded video flows to the WAN interface.
In a preferred embodiment, the method of the present invention further comprises the step of periodically updating the forwarding rules.

Short Description of the Drawings

The invention will be further elucidated by means of the following description and the appended figure.
Fig. 1 represents a schematic the different blocks of the networking device of the present invention.

Detailed Description of the Invention

The present invention will be described with respect to particular embodiments and with reference to certain drawings. The invention is however not limited thereby but only by the claims. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn on scale for illustrative purposes.
Detection of services and proper assignment of services to network segments remains a delicate task. To optimize transmission of incoming data over the different available network segments, more sophisticated rules for forwarding are necessary than pure bridging or routing. On the one hand, the different subnet segments may have different profiles in terms of for instance data capacity, latency, and packet loss. On the other hand, different services offered by the access link (e.g. WAN link) may have different requirements related to for instance capacity, latency, and packet loss.
An object of this invention is to match the requirements of the different services offered by the WAN link to the capabilities of the different subnet segments, taking into account the topological restrictions. In particular, the available network topology may provide several paths over different subnet segments from the networking device (e.g. Residential Gateway) to a first end point, while providing only one path over one specific network segment from the networking device to a second end point. Matching services to adequate network segments leads to better utilization of the available bandwidth and to a better Quality of Experience for the end user.
The present invention discloses a networking device for exchanging data frames over an interface and/or different subnets. The interface is preferably an access interface coupled for example to a public Wide Area Network (WAN) infrastructure such as the Internet. The device of the present invention is adapted to exchange data frames according to rules for forwarding data frames over the network and the available subnets.
A subnet of the present invention may be a Local Area Network (LAN) comprising different segments. A subnet may comprise native LAN segments (e.g. Ethernet, WiFi and other segments known to the person skilled in the art), as well as with segments built on legacy cabling (e.g. HPNA, MoCA and other segments known to the person skilled in the art).
A networking device is presented wherein a QoS monitoring agent is adapted to retrieve Quality of Service information, such as capacity, packet loss, and latency characteristics, from the in-house network segments. This Quality of Service information is used to dynamically update the classification rules.
Fig.1 shows the networking device of the present invention. In a preferred embodiment, the networking device (1), operating as a residential gateway, is connected to the Internet via its WAN interface (2) coupled to a PON link (3). For example, as in a typical in-house network a first subnet interface is coupled to a wireline-based network and a second subnet interface is coupled to a wireless-based network. In particular, the networking device provides access to a local IEEE Std 802.11g wireless network via its first LAN interface (4) and to a local 10/100BASE-T Ethernet network via its second LAN interface (5). The initial classification rules in the memory (6) are set to identify different video flows flows along with their minimum bit rate and maximum packet loss and latency requirements. Upon initial entry into service, the Quality of Service monitoring agent (9) queries the management information base (MIB) of both LAN interfaces (4) and (5) to obtain values of negotiated bit rates, packet loss statistics, and (inferred) link latencies. The classification rules are now extended with rules for forwarding that ensure proper attribution of the (video) flows to the LAN segment that best meets their respective requirements. From this point on, incoming packets that are recognized by the classification agent (7) as belonging to a particular flow, based on the extracted information (8), are forwarded to the most appropriate LAN. The querying of the MIB and update of the forwarding rules are repeated periodically.
In an alternative embodiment, the networking device (1), functioning as a traffic aggregator, is connected to a video decoding unit via its WAN interface (2) using a chosen point-to-point link technology (3). The traffic aggregator exchanges data with an IEEE Std 802.11g wireless network via its first LAN interface (4) and with a local 10/100BASE-T Ethernet network via its second LAN interface (5). The initial classification rules in the memory (6) are set to identify incoming coded video flows, emanating from a video server, arriving via the LAN interfaces, and video return channel information emanating from the video decoder, arriving via the WAN interface. Upon initial entry into service, the Quality of Service monitoring agent (9) queries the management information base of both LAN interfaces (4) and (5) to obtain values of negotiated bit rates, packet loss statistics, and (inferred) link latencies. The classification rules are now extended with rules for forwarding that ensure proper attribution of video return channel information to the LAN segment that is best suited to carry the information back to the video server. The classification rules are further extended with rules for forwarding subscribed coded video flows to the WAN interface. From this point on, incoming packets that are recognized by the classification agent (7) as belonging to a particular flow, based on the extracted information (8), are forwarded to the appropriate network interface. The querying of the MIB and update of the forwarding rules are repeated periodically.
In another aspect of the invention transmissions can be relayed from a first end point connected to at least one of the Residential Gateway's network segments to either a second end point connected to at least one of the Residential Gateway's network segments, or the WAN interface, taking into account the topological restrictions. This aspect is also achieved by the operation of the QoS monitoring agent described above, with the additional requirement that the classification table contain rules that distinguish local area traffic from traffic that leaves the local area network segments.. In an exemplary embodiment for this object, the networking device (1), functioning as a residential gateway, is connected to the Internet via its WAN interface (2) coupled to a PON link (3). The networking device provides access to a local IEEE Std 802.11g wireless network via its first LAN interface (4) and to a local 10/100BASE-T Ethernet network via its second LAN interface (5). The initial classification rules in memory (6) are set to identify flows from handheld data acquisition/generation devices connected to the 802.11g wireless network along with their minimum bit rate and maximum packet loss and latency requirements, said flows being destined for storage devices attached to the LAN interfaces. Upon initial entry into service, the Quality of Service monitoring agent (9) queries the management information base of both LAN interfaces (4) and (5) to obtain values of negotiated bit rates, packet loss statistics, and (inferred) link latencies. The classification rules are now extended with forwarding rules that ensure proper attribution of the flows to the LAN segment that best meets their respective requirements. From this point on, incoming packets that are recognized by the classification agent (7) as belonging to a particular flow, based on the extracted information (8), will be forwarded to the most appropriate LAN. The querying of the MIB and update of the forwarding rules are repeated periodically.
An exemplary mode for carrying out the invention is described below. IEEE Std 802.3-2005, known as the "Ethernet standard", contains specifications of several physical layer types that are commonly used in home networks, such as 10BASE-T, 100BASE-T, and 1000BASE-T. These three port types all operate over multiple strands of twisted copper pair, and employ the same physical medium connector (of the type RJ-45). The end-user can essentially consider this as one single technology, because the stations at both ends of the wire determine their best common capabilities through the auto-negotiation process. The auto-negotiation process results among other things in a common transmission rate and duplicity. IEEE Std 802.3-2005 Clause 22 specifies the physical interface between the Ethernet MAC and PHY, which consist of a data interface (MII) and a management interface (MDIO). Via registers 0.6 and 0.13 of the MDIO, the MAC can access the actual transmission rate of the PHY, while register 0.8 indicates the duplicity. The MAC can also access various error counters and ascertain whether the attached Ethernet PHY is operational or not.
Clauses 61-63 of the same standard specify the technology known as "Ethernet in the First Mile" over point-to-point voice-grade copper wiring. Although the standard situates the use of this technology in subscriber access networks, the use of 10PASS-TS in large LANs over existing wiring is not precluded. In 10PASS-TS, as in most xDSL technologies, the physical layer trains to the best achievable bit rate and latency that can be supported by the medium (barring more stringent constraints from the network manager). For 10PASS-TS, this rate can be anywhere between 0 and 100 Mb/s. The extended MDIO specified in Clause 45 of IEEE Std 802.3-2005 provides the MAC access to the actual trained speed through its register 1.31.15:5. The MAC can also access various error counters and physical transmission parameters, and ascertain whether the attached 10PASS-TS PHY is operational or not.
In Wireless LANs according to the IEEE Std 802.11-2003 standards family, management frames may contain a "Supported Rates element" (clause 7.3.2.2). Furthermore, the PHY service interface provides a DATARATE parameter in the TXVECTOR and RXVECTOR vectors (12.3.4.4). These elements allow the IEEE 802.11 MAC to learn, via the implementation of the abstract MLME_PLME_SAP interface, the theoretical and the effective data rate of the wireless medium towards various peers at any time. The IEEE 802.11 MIB contains a Dot11CountersEntry, which stores information with relevance to QoS that can be supported, such as counters of failures, retries, RTS misses, ACK misses, etc. The extensions defined by IEEE Std 802.11e additionally provide the very useful managed objects dot11QosOptionImplemented, ot11QosDiscardedFragmentCount, and dot11QosCounters. All this clearly demonstrates that the higher layers can obtain valuable QoS information via existing interfaces. The MAC can also access various error counters and physical transmission parameters, and ascertain whether the attached 802.11 PHY is operational or not.
The PHY specified in ITU-T Recommendation G.9954 on provides transmission and reception of physical layer frames over in-house phone wire media. It supports 2, 4, 8, 16 and 24 Mbaud symbol rates with 2 to 10 bits-per-symbol constellation encoding, providing data rates in the range of 4-240 Mbit/s. Again, G.9954 management includes all the facilities that are required in order to collect information from the PHY and other (sub)layers.
Based on the information collected by the MAC from the PHY, as indicated in the examples above, appropriate allocation of services to PHYs can take place. Obviously, PHYs that are no longer operational or have retrained to extremely low-quality operational parameters should not be selected for sensitive data transmissions. However, more subtle traffic allocation criteria are needed, as it is generally known that different services put different demands on the physical layer. In ITU-T G.993.1 VDSL for example, which supports the option of "dual latency", it is stated that "the lowest level of protection is required to support latency sensitive services such as voice, while the highest level is required to support burst error sensitive services such as entertainment video." As stated before, it is known in the art that the required demultiplexing can take place automatically at the level of the MAC ( EP 1328091 A ). The present invention generalizes this approach by taking into account multiple, heterogeneous PHYs, and using information that is retrieved in real time from the various PHYs via their respective management interfaces to optimize PHY selection.

Claims

A networking device (1) comprising
- an interface (2) for exchanging data frames over an access network,

- at least a first and a second subnet interface (4,5) for exchanging data frames, said subnet interfaces (4,5) arranged for being coupled to at least a first and a second network, respectively,

- a memory (6) for storing classification rules applicable to said data frames,

- a classification agent (7) coupled to said memory (6) and arranged for extracting information from an incoming data frame and for applying said classification rules to said extracted information in order to determine the interface via which said incoming data frame is to be forwarded,

- a Quality of Service monitoring agent (9) arranged for retrieving Quality of Service information from said subnet interfaces (4,5) and for dynamically updating at least one of said classification rules according to said QoS information.
The networking device of claim 1 characterised in that said device is a Residential Gateway.
The networking device of any of the previous claims characterised in that said interface (2) is a WAN interface arranged for being coupled to an access link (3).
The networking device of claim 1 or 2 characterised in that said interface (2) is a WAN interface arranged for being coupled to a point-to-point link (3).
The networking device of any of the previous claims characterised in that said first and second subnet interface (4,5) are selected from the group of interfaces consisting of {LAN, wireless LAN, Ethernet, power line, MOCA and HPNA}.
The networking device of any of the previous claims characterised in that said extracted information comprises a source MAC address and/or a destination MAC address.
The networking device of any of the claims 1 to 6 characterised in that said extracted information comprises a source IP address and/or a destination IP address.
The networking device of any of the previous claims characterised in that said extracted information comprises a virtual LAN identifier.
The networking device of any of the previous claims characterised in that said extracted information comprises a user priority indicator.
A method for exchanging data between an interface (2) for exchanging data frames over an access network and at least a first and a second subnet interface (4,5) comprising the steps of:
- setting initial classification rules for identifying data frames;

- querying said subnet interfaces (4,5) in order to retrieve QoS information;

- updating at least one of said initial classification rules to rules for forwarding data frames; and

- selecting an interface (2,4,5) according to said classification rules for forwarding an incoming data frame in said classification agent.
The method according to claim 10 characterised in that said step of updating further comprises the step of extending said rules with forwarding rules for forwarding flows to said interface (2).
The method according to any of claim 10 or 11, further comprising the step of periodically updating said rules for forwarding.